yattoite

Slack bot that responds on my behalf on holidays using freee 人事労務.

This bot is created using Slack past messages and markov chain.

You can also use a CSV file instead of Slack comments.

Requirements

• Slack
• freee API
• Docker
• docker-compose
• Heroku
  • Not required if run on local
• CSV file
  • Not required if use Slack comments

Expected CSV format

id	body
1	私は天才エンジニアです。
2	お腹空き空き空き空き空きっ腹。寿司食うぞ。
...	...

Usage

1. Prepare code and create .env file

git clone https://github.com/Doarakko/yattoite
cd yattoite
cp .env.example .env

2. Create Slack app and enter environmental varialbes to .env

Easy to create using manifest.yml.

Get App-Level Token(xapp-aaaa) with connections:write and Bot User OAuth Token(xoxb-bbbb).

SLACK_APP_TOKEN=xapp-aaaa
SLACK_BOT_TOKEN=xoxb-bbbb
SLACK_USER_ID=ABCD01234

Get your user id on Slack by tapping your account and pressing "Copy link".

3. Create freee App

FREEE_CLIENT_ID=cccc
FREEE_CLIENT_SECRET=dddd
FREEE_CODE=eeee

For FREEE_CODE, enter the "認可コード" obtained by accessing the "Web アプリ認証用 URL" on the browser.

4. Get Slack comments

docker-compose up
docker exec -it app python scripts/create_table.py

q specifies the name of the channels from which to get the comments.

All channels containing q are included. If you specify times, you can get it from *times*.

docker exec -it app python scripts/1_save_slack_comments.py <query>

Choose a user to create a bot. User id is like U0123abcd.

docker exec -it app python scripts/2_select_slack_user.py <user id>

If you use the CSV file you have already prepared, skip it.

5. Preprocessing and make text file

You can specify the input and output file path by specifying the -i and -o options.(optional)

docker exec -it app python scripts/3_preprocessing.py -i data/<user id>_comments.csv

You can use -min and -max options to limit the size of the sentence you use.(optional)

If the -min is 6, then 我々4歳くらいです(我々 4 歳 くらい です) is excluded.

docker exec -it app python scripts/4_make_sentences.py -i <input> -o <output> -min <minimum word count> -max <max word count>

6. Generate sentences

You can use -s to specify how many morphemes of the original sentence to use.

Larger will produce the correct sentece, but it's boring.

docker exec -it app python scripts/5_generate.py -i <input> -o <output> -n <count of generate sentences> -s <state size>

Let's enjoy tuning!

7. Insert records

docker exec -it app python scripts/6_insert.py

Run on Heroku

Click "Deploy to Heroku" Button and enter your environment valiables.

Get DATABASE_URL from "Settings/Config Vars" on Heroku dashboard.

On local enter the DATABASE_URL, other data create and insert methods are the same as the above local step.

Hints

Bot does not respond when you are working

def message(message):
    ...

    if freee.is_working():
        return

    ...
    client.chat_postMessage(channel=message["channel"], text=row.body)

Make a rug without getting an immediate response

def message(say):
    ...

    # camouflage to avoid being identified as a bot
    time.sleep(
        random.choices(
            [10, 20, 30, 60, 120, 180, 240, 300, 360, 420, 480, 540, 600],
            [3, 3, 3, 5, 5, 5, 3, 3, 1, 1, 1, 1, 1],
        )[0]
    )

    ...

Make your bot's icon and name the same as your Slack account

If you look closely, it will be noticed if it is a bot depending on whether or not "App" is displayed.

Well, it'll be alright.